import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

import seaborn as sn

raw_data= pd.read_csv("vehicle-1.csv")

raw_data.head()

raw_data.corr()

raw_data.describe()

sn.pairplot(raw_data, diag_kind = 'kde')

plt.show()

raw_data['class']= raw_data['class'].replace({'van': 1, 'bus': 2, 'car': 3})

​

raw_data.describe()

data= raw_data.drop(['class'], axis= 1)

data.median()

data.fillna(data.median(), inplace= True)

max_length_aspect_ratio= data['max.length_aspect_ratio']

sn.boxplot(data= max_length_aspect_ratio)

​

for i in range(846):

    if max_length_aspect_ratio[i]> 13:

        max_length_aspect_ratio[i]= max_length_aspect_ratio.median()

​

data['max.length_aspect_ratio']= max_length_aspect_ratio

pr_access_aspect_ratio= data['pr.axis_aspect_ratio']

sn.boxplot(data= pr_access_aspect_ratio)

​

for i in range(846):

    if pr_access_aspect_ratio[i]> 80:

        pr_access_aspect_ratio[i]= pr_access_aspect_ratio.median()

​

data['pr.axis_aspect_ratio']= pr_access_aspect_ratio

​

radius_ratio= data['radius_ratio']

sn.boxplot(data= radius_ratio)

​

for i in range(846):

    if radius_ratio[i]> 250:

        radius_ratio[i]= radius_ratio.median()

​

​

data['radius_ratio']= radius_ratio

data.describe()

sn.pairplot(data, diag_kind = 'kde')

plt.show()

from sklearn.preprocessing import StandardScaler

data = StandardScaler().fit_transform(data)

from sklearn.model_selection import train_test_split

x_train, x_test, y_train, y_test= train_test_split(data, raw_data['class'], test_size=0.3, random_state=0)

from sklearn import svm

svma= svm.SVC()

svma.fit(x_train, y_train)

svm_prediction= svma.predict(x_test)

from sklearn.metrics import classification_report

from sklearn import metrics

print(classification_report(y_test,svm_prediction))

print("Accuracy:",metrics.accuracy_score(y_test, svm_prediction))

metrics.confusion_matrix(y_test, svm_prediction)

from sklearn.model_selection import cross_val_score

clf = svm.SVC(kernel='linear', C=1)

scores = cross_val_score(clf, data, raw_data['class'], cv=5)

scores    

print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))

# Principal Component Analysis

from numpy import array

from sklearn.decomposition import PCA

# create the PCA instance

pca = PCA(8)

# fit on data

pca.fit(data)

print(pca.components_)

print(pca.explained_variance_)

# transform data

pcadata = pca.transform(data)

print(pcadata)

x_train, x_test, y_train, y_test= train_test_split(pcadata, raw_data['class'], test_size=0.3, random_state=0)

svma= svm.SVC()

svma.fit(x_train, y_train)

svm_prediction= svma.predict(x_test)

print(classification_report(y_test,svm_prediction))

print("Accuracy:",metrics.accuracy_score(y_test, svm_prediction))

metrics.confusion_matrix(y_test, svm_prediction)

from sklearn.model_selection import cross_val_score

clf = svm.SVC(kernel='linear', C=1)

scores = cross_val_score(clf, data, raw_data['class'], cv=5)

scores    

print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))